index1-file.js

// 1. Import necessary modules and libraries
import { OpenAI } from 'langchain/llms';
import { RetrievalQAChain } from 'langchain/chains';
import { HNSWLib } from 'langchain/vectorstores';
import { OpenAIEmbeddings } from 'langchain/embeddings';
import { RecursiveCharacterTextSplitter } from 'langchain/text_splitter';
import * as fs from 'fs';
import * as dotenv from 'dotenv';

// 2. Load environment variables
dotenv.config();

// 3. Set up input data and paths
const txtFilename = "wi320";
const question = "Tell me about WI-320?";
const txtPath = `./${txtFilename}.txt`;
const VECTOR_STORE_PATH = `${txtFilename}.index`;

// 4. Define the main function runWithEmbeddings
export const runWithEmbeddings = async () => {
  // 5. Initialize the OpenAI model with an empty configuration object
  const model = new OpenAI({});

  // 6. Check if the vector store file exists
  let vectorStore;
  if (fs.existsSync(VECTOR_STORE_PATH)) {
    // 6.1. If the vector store file exists, load it into memory
    console.log('Vector Exists..');
    vectorStore = await HNSWLib.load(VECTOR_STORE_PATH, new OpenAIEmbeddings());
  } else {
    // 6.2. If the vector store file doesn't exist, create it
    // 6.2.1. Read the input text file
    const text = fs.readFileSync(txtPath, 'utf8');
    // 6.2.2. Create a RecursiveCharacterTextSplitter with a specified chunk size
    const textSplitter = new RecursiveCharacterTextSplitter({ chunkSize: 1000 });
    // 6.2.3. Split the input text into documents
    const docs = await textSplitter.createDocuments([text]);
    // 6.2.4. Create a new vector store from the documents using OpenAIEmbeddings
    vectorStore = await HNSWLib.fromDocuments(docs, new OpenAIEmbeddings());
    // 6.2.5. Save the vector store to a file
    await vectorStore.save(VECTOR_STORE_PATH);
  }

  // 7. Create a RetrievalQAChain by passing the initialized OpenAI model and the vector store retriever
  const chain = RetrievalQAChain.fromLLM(model, vectorStore.asRetriever());

  // 8. Call the RetrievalQAChain with the input question, and store the result in the 'res' variable
  const res = await chain.call({
    query: question,
  });

  // 9. Log the result to the console
  console.log({ res });
};

// 10. Execute the main function runWithEmbeddings
runWithEmbeddings();